Biomedical Engineering Reference
In-Depth Information
Fig. 6.6
Initial figure (black) and described and inscribed figures (horizontal lines and gray)
Fig. 6.7
The construction of the inscribed and circumscribing figures
inscribed into the initial image. Then the areas of initial and inscribed figures are
calculated. The difference between them is used as the first feature for shape
recognition.
Since the neurocomputer is oriented to work with the binary vectors of large
dimensionality, the algorithm is built in such a way that, simultaneously, as many
circles as pixels are counted in one image line (512 pixels in the last version of the
neurocomputer). The extraction and the filling of cavities on the complex figure are
done with the aid of the procedure of scanning by circle. The procedure must be
repeated for the initial image. When a quantity of 1s in the circle exceeds a certain
threshold (
P
= 13), the center of the circle is included in the figure (Fig.
6.7
, case 2).
Through several iterations, the initial image will be circumscribed by a certain
figure whose points will fill its cavities. After calculating the areas of the initial and
circumscribing figures and determining their difference, we will obtain the second
feature for shape recognition. The processes of the truncation of protrusions and the
